Ludwig Nussel wrote:
I'm just looking through this for the next VDR version.
What's the meaning of this? I can't see any place where VFALDT is actually used.
Is it really necessary to have VDR_USER _and_ VDR_GROUP separately? I'd say VDR_USER should suffice.
I assume this means there are systems that don't provide this. Is there a runtime method to determine the presence of this?
Am I missing something here, or is the su() call always done, no matter under which user ID VDR has been started? Shouldn't this only be done if it was started as 'root'?
Klaus
* Klaus Schmidinger schrieb am 30.12.05, um 18:19 Uhr:
Thats really nice to hear.
For me this looks like a typo by accident. ;-)
Well, you need libcap to be able to use this patch.
You are completely right, these calls should only be done when root is calling vdr.
I attached the patch which debian and ctvdr use since allmoast one year without bigger drawbacks. (it is based on Ludwig Nussel's patch, but sligtly modified)
Regards, Thomas
Klaus Schmidinger wrote:
Looks like garbage :-)
I actually don't use it myself either. The primary+supplementary groups of the user are used then.
Well, you need the definitions at compile time. It's not really necessary to use libcap though, the interface is clumsy anyways. You could do the syscall yourself as well. I just didn't bother changing it as noone requested it and libcap works for me :-)
That was basically the case when I used command line switches. I had problems with some plugins (don't remember which exactly) as they already created some fixed files as the invoking use e.g. when you ran --help and then prevented start of vdr later. Yeah, all horrible. Maybe that kind of things get sorted if you include the patch though.
cu Ludwig